Methods for studying the pareto-fronts in multi-objective design optimization problems of electrical machines

The paper describes a generic multi-objective problem for surface mounted permanent magnet rotor synchronous machines operated from current-regulated sine-wave power electronic drives. Three objectives are concurrently considered for the maximum specific torque per mass of PM, the minimum specific losses per mass of winding copper and laminated steel, and the minimum relative torque ripple as a fraction of rated average torque. These formulations, together with the use of a parametric model with eight input variables defined relatively to a ninth independent variable, the air-gap (stator inner) diameter, which is the only one employing absolute dimensions, ensure the generality of the method and of the results in a wide range of rated power from 1kW to 1MW. A Differential Evolution optimization algorithm with 4,800 candidate designs, analyzed through a computational efficient electromagnetic FEA technique, is employed. Procedures for systematically studying the resultant optimal pareto-front are proposed. The reported findings include insights into system correlations and engineering recommendations for selecting optimal values for the design input variables.